Sign, derivative, and cancellation audit

The source derivative is ambient and precedes incidence restriction. The derivative in row $6$ is line-only.

Every two-line kernel is differentiated in independent seed coordinates before restriction to its cones.

The internal $5z$ identity transfers two $r$ derivatives, producing the signs $+,-2,+$.

Frozen coefficient derivatives and total derivatives of the parameter-dependent amplitude are distinguished in the boundary decomposition.

The temporal boundary density is ${\mathrm{d}}^3\mathbf{s}$. It is not replaced by $\mathrm{d}R(s)$ without the corresponding factor $4\pi q$.

The first-correction proof moves the translated temporal boundary to a fixed integration domain before differentiating.

The curvature correction retains $|{\mathbf{H}}_{\perp }{|}^2/(2q)$. Radial integration by parts fixes its sign through $B_{\perp }=-\mathcal{L}$.

The operator $\mathcal{C}=4\mathcal{A}+3\mathcal{B}$ includes the first correction needed for the fourth master derivative. An unspecified $O({\delta }^{-3})$ error would not determine it.

The spatial propagation derivatives contain tangential derivatives of $\mathcal{G}$. Their sum is $\mathcal{V}[\varphi ]$; omitting this term would give an incorrect $P_3$.

The mass vertex contributes $\kappa \mathcal{G}[\varphi ]/2$ to $P_3$. It is lower order in the earlier $D^{2}M$ calculation, but not in the $5z$ coefficient.

The scale derivative is checked with the differentiated profile on both boundary and interior terms. No error inequality is differentiated.

The order-seventeen seminorm follows from the first-correction derivative budget and two incidence spectators. It is sufficient, not claimed optimal.

The $5a$–response identity uses ${\square }_{\eta }R={\delta }^{(4)}$ at the inner vertex. It is not an evaluation of a separated-stratum formula at $r=0$.

Explicit cancellation of an incidence term does not annihilate the remaining spatial-Laplacian distribution.

The raw obstruction is a fixed separated test and remains independent of the signed $5z$ asymptotics.

Necessary correction support is not sufficient correction existence.

The leading remainder estimate does not establish either divergence or convergence of the leading-subtracted family.

No registry statement, simulated quadrature, physical analogy, or extension ambiguity is used as an analytic premise.